Analysis of data from clinical studies of localized juvenile periodontitis.

Data from 27 subjects with localized juvenile periodontitis were used to compute estimates of the intraclass correlation coefficient for initial periodontal pocket depth measurements and total subgingival Actinobacillus actinomycetemcomitans counts, and for changes from pre- to post-treatment in periodontal pocket depth, probing attachment level and Actinobacillus actinomycetemcomitans counts. The estimates for the clinical parameters were considerably higher than those found in a similar study of adult periodontitis. On the average, the estimates for the intraclass correlation coefficients for Actinobacillus actinomycetemcomitans counts did not differ markedly from the corresponding estimates for the clinical parameters. Although the results of this study indicate that the major component of variability in parameters commonly used in studies of periodontal disease can be attributed to site-specific factors, sites within subjects cannot be considered statistically independent. For the analysis of data arising from studies of periodontal disease, it is recommended that statistical techniques, like the nested mixed effects analysis of variance model be employed, which account for this dependence. These techniques will allow meaningful biological questions concerning site-specific phenomenon to be answered. However, changes in clinical and microbiological parameters, when therapeutic measures are employed on localized juvenile periodontitis patients, are often dramatic enough to allow informal data presentation.